AVLEN_supplementary_doc.pdf: Main supplementary document
code folder: contains four python script that covers 
qualitative.zip: contains the example videos and a presentation slide with the example videos


Please use a suitible media player (e.g., VLC media player) to hear the sound. The sound is not audible in Powerpoint. 


Following are the videos (in ./qualitative/videos/ subfolder) that uses AVLEN to query instruction and navigate towards the target sounding object:
(Whenever agent queries and receives natural languages instruction it is shown on that particular frame.)
 
'fzynW3qQPVF_22767_cabinet_spl0.31.mp4': video of an agent following sound coming from cabinet using AVLEN. 
'jtcxE69GiFV_2648_cabinet_spl1.00.mp4': video of an agent following sound coming from cabinet using AVLEN.
'pa4otMbVnkk_14394_picture_spl0.14.mp4': video of an agent following sound coming from picture using AVLEN.
'pa4otMbVnkk_15330_picture_spl1.00.mp4': video of an agent following sound coming from picture using AVLEN.
'pa4otMbVnkk_22608_cushion_spl0.88.mp4' video of an agent following sound coming from cushion using AVLEN.


In addition we have provided two more videos for scene 'pa4otMbVnkk' and episode '15330':

'pa4otMbVnkk_15330_picture_spl0.00_savi.mp4': video of an agent following sound coming from picture using only audio-goal policy (\pi_g).
'pa4otMbVnkk_15330_picture_spl0.00_jask.mp4': video of an agent following sound coming from picture using Model Uncertainty based query selection approach (MU)